"undefined method on_booted for nil:NilClass" on "pumactl start"

[opti]

Describe the bug
Having on_booted event in puma configuration file fails to start puma server using pumactl -F config/puma.rb start.

It raises error here since @config.options[:events] is nil

puma/lib/puma/dsl.rb

Lines 706 to 708 in 5fc43d7

	def on_booted(&block)
	@config.options[:events].on_booted(&block)
	end

while trying to load config here:

puma/lib/puma/control_cli.rb

Lines 130 to 131 in 5fc43d7

	config = Puma::Configuration.new({ config_files: [@config_file] }, {})
	config.load

Puma config:

on_booted do
  puts "on_booted called"
end

To Reproduce

Copy this into a file called hello.ru:

run lambda { |env| [200, {"Content-Type" => "text/plain"}, ["Hello World"]] }

Create config file called puma.rb with the following content:

on_booted do
  puts "on_booted called"
end

Run it with:

bundle exec pumactl -F puma.rb start hello.ru

Observe error:

undefined method `on_booted' for nil:NilClass

The same can be reproduced with test config in the repository:

bundle exec pumactl -F test/config/event_on_booted.rb start test/rackup/hello.ru

Expected behavior
Puma server should start

Desktop:

• OS: macOS 14.4.1
• Puma Version 6.4.2
• ruby 3.2.3

[dentarg]

One-liner to reproduce:

$ echo 'app { [200, {}, ["OK"]] }; on_booted { puts "puma booted" }' | ruby -Ilib bin/pumactl --config-file /dev/stdin start
undefined method `on_booted' for nil:NilClass

I guess #2709 never tested with pumactl

Just to be clear, the hook works using puma

$ echo 'app { [200, {}, ["OK"]] }; on_booted { puts "puma booted" }' | ruby -Ilib bin/puma --config /dev/stdin --port 0 --log-requests
Puma starting in single mode...
* Puma version: 6.4.2 (ruby 3.2.3-p157) ("The Eagle of Durango")
*  Min threads: 0
*  Max threads: 5
*  Environment: development
*          PID: 26832
* Listening on http://0.0.0.0:61890
Use Ctrl-C to stop
puma booted
^C- Gracefully stopping, waiting for requests to finish
=== puma shutdown: 2024-03-29 09:46:32 +0100 ===
- Goodbye!

[dentarg]

Looks like events is missing at

puma/lib/puma/control_cli.rb

Line 130 in 2314d7a

config = Puma::Configuration.new({ config_files: [@config_file] }, {})

as a61b078 added that in several places

However, wouldn't it be better to make events part of the default configuration? Is there anything preventing us from that?

puma/lib/puma/configuration.rb

Lines 127 to 173 in 2314d7a

	class Configuration
	DEFAULTS = {
	auto_trim_time: 30,
	binds: ['tcp://0.0.0.0:9292'.freeze],
	clean_thread_locals: false,
	debug: false,
	early_hints: nil,
	environment: 'development'.freeze,
	# Number of seconds to wait until we get the first data for the request.
	first_data_timeout: 30,
	# Number of seconds to wait until the next request before shutting down.
	idle_timeout: nil,
	io_selector_backend: :auto,
	log_requests: false,
	logger: STDOUT,
	# How many requests to attempt inline before sending a client back to
	# the reactor to be subject to normal ordering. The idea here is that
	# we amortize the cost of going back to the reactor for a well behaved
	# but very "greedy" client across 10 requests. This prevents a not
	# well behaved client from monopolizing the thread forever.
	max_fast_inline: 10,
	max_threads: Puma.mri? ? 5 : 16,
	min_threads: 0,
	mode: :http,
	mutate_stdout_and_stderr_to_sync_on_write: true,
	out_of_band: [],
	# Number of seconds for another request within a persistent session.
	persistent_timeout: 20,
	queue_requests: true,
	rackup: 'config.ru'.freeze,
	raise_exception_on_sigterm: true,
	reaping_time: 1,
	remote_address: :socket,
	silence_single_worker_warning: false,
	silence_fork_callback_warning: false,
	tag: File.basename(Dir.getwd),
	tcp_host: '0.0.0.0'.freeze,
	tcp_port: 9292,
	wait_for_less_busy_worker: 0.005,
	worker_boot_timeout: 60,
	worker_check_interval: 5,
	worker_culling_strategy: :youngest,
	worker_shutdown_timeout: 30,
	worker_timeout: 60,
	workers: 0,
	http_content_length_limit: nil
	}